The long road of hypertext
Hypertext is older than the web. In 1945 Vannevar Bush described the Memex, a desk that stored a person's books and notes and let them link one passage to another.[1] Nothing like it could be built then, but the idea of following a trail of associations instead of reading in a straight line stayed alive in a small circle of engineers and writers.
Ted Nelson coined the word hypertext in 1963 and spent the next decades building Xanadu, a system in which every quotation would stay linked to its source. Xanadu was never finished in the way he imagined. Still, the web borrowed its most visible trick, the link, and left behind most of the ambition about credit and versioning.
In 1967 Nelson also described stretchtext, text that grows where the reader wants more and shrinks back where they do not. A summary could unfold into a paragraph, and a paragraph into a chapter, all in the same place on the page. The reader would decide how deep to go, sentence by sentence.
The trouble was labor. To offer stretchtext, a writer had to produce every level by hand, and keep the levels consistent whenever the text changed. Almost nobody had the time, so the experiments stayed small: a few hypertext fiction pieces, some teaching tools, and an occasional outline view in a word processor.
Language models change the cost. A short version of any sentence can now be generated in a moment, while the original stays exactly as its author wrote it. That is the arrangement this extension uses: the short phrase is new, and everything that appears when you click it is the page's own text, in place and with its links intact.
The idea fits readers who skim for a living. Most of what lands in front of a person on a given day is not worth a full read, but some paragraphs are exactly what they came for. Stretchtext lets them tell which is which with one click and no scrolling, and then go back to skimming.